[RFA] rs6000-tdep.c: figure out e500 machine from bfd.

[RFA] rs6000-tdep.c: figure out e500 machine from bfd.

[Elena Zannoni]

A powerpc executable doesn't have indications of which variant it is
meant to run on. The design choice was that in theory everything
should run on any version of the processor.  For gdb's purposes we
have now a new section that is specific to e500, which can help us
identify the executable and set all the architecture dependents bits
in the correct way.

Elena
[sorry Kevin, it's almost over...:-)]


2002-08-20  Elena Zannoni  <ezannoni@redhat.com>

	* rs6000-tdep.c (rs6000_gdbarch_init): Figure out whether we have
	an e500 executable.



Index: rs6000-tdep.c
===================================================================
RCS file: /cvs/uberbaum/gdb/rs6000-tdep.c,v
retrieving revision 1.75
diff -u -p -r1.75 rs6000-tdep.c
--- rs6000-tdep.c	20 Aug 2002 22:24:29 -0000	1.75
+++ rs6000-tdep.c	20 Aug 2002 22:36:03 -0000
@@ -2426,6 +2426,7 @@ rs6000_gdbarch_init (struct gdbarch_info
   bfd abfd;
   int sysv_abi;
   enum gdb_osabi osabi = GDB_OSABI_UNKNOWN;
+  asection *sect;
 
   from_xcoff_exec = info.abfd && info.abfd->format == bfd_object &&
     bfd_get_flavour (info.abfd) == bfd_target_xcoff_flavour;
@@ -2499,6 +2500,20 @@ rs6000_gdbarch_init (struct gdbarch_info
   tdep = xmalloc (sizeof (struct gdbarch_tdep));
   tdep->wordsize = wordsize;
   tdep->osabi = osabi;
+
+  /* For e500 executables, the apuinfo section is of help here.  */
+  if (info.abfd)
+    {
+      sect = bfd_get_section_by_name (info.abfd, ".PPC.EMB.apuinfo");
+      if (sect)
+	{
+	  arch = info.bfd_arch_info->arch;
+	  mach = bfd_mach_ppc_e500;
+	  bfd_default_set_arch_mach (&abfd, arch, mach);
+	  info.bfd_arch_info = bfd_get_arch_info (&abfd);
+	}
+    }
+
   gdbarch = gdbarch_alloc (&info, tdep);
   power = arch == bfd_arch_rs6000;

Re: [RFA] rs6000-tdep.c: figure out e500 machine from bfd.

[Kevin Buettner]

On Aug 20,  6:36pm, Elena Zannoni wrote:

> +  /* For e500 executables, the apuinfo section is of help here.  */
> +  if (info.abfd)
> +    {
> +      sect = bfd_get_section_by_name (info.abfd, ".PPC.EMB.apuinfo");
> +      if (sect)

What is the apuinfo section?  If you know what it's about (I sure
don't), consider expanding the comment somewhat.

Is there any chance that this section could appear in non-e500
executables?  If so, make a note of this gotcha in the comment too.

Otherwise, fine.  Go ahead and commit it after expanding the comment
as necessary.

Kevin

Re: [RFA] rs6000-tdep.c: figure out e500 machine from bfd.

[Elena Zannoni]

Kevin Buettner writes:
 > On Aug 20,  6:36pm, Elena Zannoni wrote:
 > 
 > > +  /* For e500 executables, the apuinfo section is of help here.  */
 > > +  if (info.abfd)
 > > +    {
 > > +      sect = bfd_get_section_by_name (info.abfd, ".PPC.EMB.apuinfo");
 > > +      if (sect)
 > 
 > What is the apuinfo section?  If you know what it's about (I sure
 > don't), consider expanding the comment somewhat.
 > 

Yes, sorry (I have been playing around with this code for too long,
obviously). It is a section that contains the version numbers of the
various Application Processing Units that the processor contains.  So
that one can detect what version/type of processor you need to run the
code on. For instance if the APU info says you need version 3 you know
that the code won't run on version 2. Because there may be new
instructions, etc. I think this info could in theory be used by the
linker, but I am not sure it is used right now.  Gas keeps track of
the rev number associated with each instruction, and then outputs (in
the apuinfo section) the highest it has found. Right now, though, it
doesn't do that. It just spews out a 1.

 > Is there any chance that this section could appear in non-e500
 > executables?  If so, make a note of this gotcha in the comment too.
 > 

No. Only e500.

 > Otherwise, fine.  Go ahead and commit it after expanding the comment
 > as necessary.
 > 

OK. Done.

 > Kevin